MemBrain/.env.example.dev at main · feelingai-team/MemBrain · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
LLM_API_URL=http://localhost:4000/v1
LLM_API_KEY=sk-1234
COMPOSE_PROJECT_NAME=mem-brain
LOCAL_DATA_DIR=./database
EXPS_DIR=./evaluation/exps

DB_HOST=localhost
DB_PORT=5432
DB_USER=postgres
DB_PASSWD=MemBrain
DB_NAME=MemBrain

# Backend server mode: dev | evaluation | demo
BACKEND_MODE=dev
BACKEND_PORT=9574
WEB_PORT=5173
ALLOWED_HOSTS=localhost,127.0.0.1

# Embedding service
# Local vLLM (default):
EMBED_SERVICE_URL=http://localhost:9113/v1/embeddings
EMBED_API_KEY=
EMBED_MODEL=qwen3-embed
EMBED_DIM=2560          # Must match model output dim; if changed, recreate the DB
# Online alternative example (OpenAI):
# EMBED_SERVICE_URL=https://api.openai.com/v1/embeddings
# EMBED_API_KEY=sk-...
# EMBED_MODEL=text-embedding-3-small
# EMBED_DIM=1536

# Rerank service
# Local vLLM (default):
RERANK_SERVICE_URL=http://localhost:9114/v1/rerank
RERANK_API_KEY=
RERANK_MODEL=qwen3-rerank
# Online alternative example (Cohere):
# RERANK_SERVICE_URL=https://api.cohere.ai/v1/rerank
# RERANK_API_KEY=...
# RERANK_MODEL=rerank-v3.5

# Fact extraction batching
EXTRACT_BATCH_MAX_MESSAGES=10
EXTRACT_BATCH_MAX_CHARS=1500
EXTRACT_CONTEXT_TAIL_SIZE=10

# Entity tree (hierarchical clustering)
TREE_MAX_CHILDREN=15
TREE_MERGE_THRESHOLD=8

# QA retrieval
QA_BM25_FACT_TOP_N=15
QA_EMBED_FACT_TOP_N=15
QA_ENTITY_TOP_N=5
QA_TREE_BEAM_WIDTH=3
QA_TREE_FACT_TOP_N=15
QA_BUDGET_MAX_TOKENS=2000
QA_MAX_PER_LEAF_ASPECT=3
QA_MAX_PER_MID_ASPECT=8
QA_RERANK_TOP_K=10
QA_MULTI_QUERY_ENABLED=true
QA_SESSION_BM25_TOP_N=10
QA_SESSION_FINAL_TOP_N=5
QA_BM25_MSG_TOP_N=5
QA_LLM_MODEL=gpt-4.1
QA_LLM_COT_MAX_TOKENS=1024

# PersonaMEM v2 virtual session splitting
PERSONAMEM_VIRTUAL_SESSION_SIZE=50